We are planning to migrate our operational DMS cluster from on premises hardware to Azure cloud. Some of the DataMiner agents will be hosting the local database locally and some will be hosting the local database on a separate server.
Which would be the recommended virtual machine series from Azure cloud for each of the 4 types of instances?
- VM hosting DataMiner SW + Cassandra Node + Elasticsearch Node
- VM hosting DataMiner SW + Elasticsearch Node
- VM hosting DataMiner SW
- VM hosting Cassandra Node or Elasticsearch Node
Jeroen
There is no easy answer to this such complex question, as there are 250+ instances that cater to different business needs. Unfortunately, the elasticity provides that challenge as the users need to take CPU, memory, disk, network bandwidth requirements along with whether it's acceptable to use the shared resources with other systems and other customers.
First key questions to explore are
- Do you want to use shared resources from Azure? Or do you want to give dedicated hosts for the DataMiner system?
- What size of DataMiner system do you want to build? Dojo will help you in defining these resource needs. Or you can use the node calculator.
- What is your budget for supporting such a cloud system?
Assuming your answers are it's a DataMiner test system with the minimum requirements, no problems with sharing resources, and flexible budget ...
For
- VM hosting DataMiner SW + Cassandra Node + Elasticsearch Node
DataMiner node calculator suggests the following for 10 elements.
So, you may want to select
- D16a v4 instance for DataMiner SW
- D16a v4 instance for Cassandra ( need to add Azure Blob to meet the disk needs, as default comes with 400 GB)
- D16a v4 instance for Elasticsearch ( need to add Azure Blob to meet the disk needs, as default comes with 400 GB)
As I mentioned, there is no single answer. The DataMiner system needs to be benchmarked against many factors for using the correct Azure instance.
In terms of what you can do on the cloud, this is just a starting point.